Regresar

Parametric analysis and comparison of models used in the analysis of steel structures

Abstract:

Performance-based seismic engineering has been gaining acceptance in the last decade. This philosophy seeks to ensure that a structure satisfies specific behavioral requirements for a given event. There are different guides and methodologies to evaluate structural performance, but conceptually they do not keep much difference. In order to perform an evaluation a mathematical model is required; however. However, the idealizations to represent the behavior of structural elements can vary significantly, giving rise to models with different degrees of complexity in their implementation or computational costs. Furthermore, the wide range of approaches, such as lumped or distributed plasticity and their calibration, can generate different results. To this, it must be added that not always a robust model is the best option for a performance analysis since it would not be practical to implement. This paper compares the pertinence of different modeling approaches to conduct a performance analysis on an 8-story building with Steel Special Moment Frames under different seismic intensity levels. Moreover, this article establishes the influence of various parameters (i.e., hysteretic detailing, Rayleigh damping, axial-moment interaction) within mathematical models on the overall response of the structure. For the purpose, FEMA P-695 methodology, as well as the ASCE 7-16 maximum-drifts requirements were used to run a sensitivity analysis between models aiming to warn of inappropriate practices. Within the findings, it can be emphasized that the secondary stiffness (in-cycle deterioration) of the element is the parameters that influence the most in the collapse response as opposed to cyclic degradation. Also, it must be noted that adding a smother yielding transition causes a higher seismic performance, which could be unconservative within an evaluation. Furthermore, the analyses illustrate how misusing an initial-stiffness proportional Rayleigh damping produces unrealistic performance values. Additional comments are made on the methodologies used to account for axial-moment interaction in lumped-plasticity models.